Expression of Placenta Growth Factor Is Associated with Unfavorable Prognosis of Advanced-Stage Serous Ovarian Cancer

Tohoku J. Exp. Med., 2018, 244PGF, is291-296 a Prognostic Biomarker in Advanced-Stage Serous Ovarian Cancer 291

Qin Meng,1 Pengjing Duan,2 Lin Li3 and Yongmei Miao3

1Department of Gynecology and Obstetrics, Shandong Medical College Linyi, Shandong, China 2Department of Gynecology and Obstetrics, Affiliated Hospital of Shandong Medical College Linyi, Shandong, China 3Department of Gynecology and Obstetrics, Linyi People’s Hospital, Linyi, Shandong, China

Ovarian cancer is the fourth leading cause of cancer death in women and the most fatal gynecologic malignancy. Placenta growth factor (PGF), a member of the vascular endothelial growth factor, plays an important role in angiogenesis. The overexpression of PGF was observed in several types of cancers, but the clinical significance of PGF in epithelial ovarian cancer (EOC) is still unknown. To explore the prognostic value of PGF among patients with serous EOC, we analyzed the expression of PGF in 89 EOC specimens by immunohistochemistry. The scoring system of immunohistochemistry was based on the staining intensity and the percentage of PGF-positive cells in each EOC tissue. According to the immunohistochemical score, 34 patients with score ≥ 6 were defined as high PGF expression, and other 55 patients were the group with low PGF expression. The prognostic significance of PGF expression was analyzed. EOC patients with higher IHC scores of PGF expression are significantly associated with positive lymphatic invasion and poorer response to chemotherapy. Patients with higher IHC scores of PGF expression had poorer response to chemotherapy and lower overall survival rate. Additionally, the positive lymph node metastasis, advanced TNM stage, and poorer response to chemotherapy were all remarkably correlated to poorer prognosis. In conclusion, patients with higher PGF in EOC tissues were more predisposed to positive lymphatic invasion, poorer response to chemotherapy and unfavorable prognosis of patients with serous EOC. We propose that PGF expression may be predictive of chemoresistance and poor prognosis of serous EOC.

of patients with chemoresistance are only 25%-35% Introduction (Colombo et al. 2014). The current standard treatment for Ovarian cancer is the fourth leading cause of cancer epithelial ovarian cancer (EOC) is the aggressive surgical death in women and the most fatal gynecologic malignancy cytoreduction followed by adjuvant chemotherapy. New with about 200,000 newly diagnosed patients and 15,000 treatment approaches and modified chemotherapies rely on deaths (Jeruss and Woodruff 2009; Zaid et al. 2013). As a the exploration of new biomarkers and drug targets. Thus, heterogeneous group of neoplasms, ovarian cancers mainly the prognostic and predictive biomarkers for EOC are still consist of malignant epithelial tumors (Cho and Shih Ie in urgent need. 2009). The histological subtypes generally include serous, Placenta growth factor (PGF), a member of the vascu- endometrioid, clear cell and mucinous adenocarcinomas lar endothelial growth factor (VEGF) sub-family, plays an (Basu et al. 2014). Unfortunately, partially because of no important role in angiogenesis and endothelial cells growth specific symptoms at the early stage, approximately 70% of (Rolny et al. 2011). PGF could be secreted out and trigger patients are present with a diagnosis of ovarian carcinoma downstream signaling pathway mainly by binding to in an advanced stage (Barnes et al. 2002). In the past VEGFR-1 or neuropilin (Rolny et al. 2011). PGF is highly decades, with the progress of modern surgeries and adju- expressed in placenta throughout all stages of gestation. As vant chemotherapies, the overall prognosis of patients with a key member of angiogenesis, PGF is expressed in endo- ovarian cancers have improved remarkably, but most thelial cells of many organs including ovary (De Falco patients would experience the relapse and the overall sur- 2012). The aberrant expression of PGF is associated with vival rate remains unsatisfactory. The 5-year survival rates many kinds of human diseases. For example, serum PGF is

Received January 15, 2018; revised and accepted March 12, 2018. Published online April 10, 2018; doi: 10.1620/tjem.244.291. Correspondence: Yongmei Miao, M.D., Department of Gynecology and Obstetrics, Linyi People’s Hospital, 27# Jiefang Road, Linyi, Shandong 264000, China. e-mail: [email protected]

291 292 Q. Meng et al. a potential biomarker for preeclampsia during pregnancy. pathologists who were blind to the clinical data. The score system of In both early and late onset preeclampsia, maternal serum IHC was dependent on the staining intensity and positive cells per- levels of PGF are lower in women presenting with pre- centage according to previous studies (Liu et al. 2017a; Xu et al. eclampsia (Khalil et al. 2008). The elevation of PGF 2017). The score of staining intensity was defined as follows: 0, neg- expression was reported in many types of cancers, includ- ative; 1, weak; 2, moderate; and 3, strong. The score of positive cells ing gastric cancer, colorectal cancer, hepatocellular carci- percentage was defined as follows: 1, < 25% of positive cells; 2, 25%-50% of positive cells; 3, 50%-75% of positive cells; and 4, 75%- noma, breast cancer and lung cancer (Dewerchin and 100% of positive cells. Thus, the final IHC score was calculated as Carmeliet 2014; Song et al. 2016). In ovarian cancer cells, the product of the score of staining intensity multiplied by the score PGF has been demonstrated to promote cell invasion via of positive cells percentage, from 0 to 12. The optimal cut-off of PGF activating zinc finger E-box-binding homeobox 2 (ZEB2) expression was identified by the point with the highest sum of sensi- and promote metastases via suppressing miR-543 and up- tivity and specificity in ROC curve, which was 5.0 in our test. The regulating MMP7 (Song et al. 2015, 2016). However, all total cohort was divided into the group with high expression of PGF the previous studies were limited in experiments in vitro, (score ≥ 6) and the group with low expression of PGF (score ≤ 4) and the clinical significance of PGF in EOC is still with the optimal cut-off. unknown. In this study, we enrolled 89 patients with serous EOC Statistical analysis and investigated the expression of PGF with immuhohisto- All the significant differences were analyzed with SPSS 22.0 chemistry (IHC). By univariate and multivariate analysis, software (SPSS, Inc., Chicago, IL, USA). The correlation between we evaluated the prognostic significance of PGF among the expression of PGF and the clinicopathological parameters was patients with serous EOC. analyzed with Fisher test. The survival curves were estimated with the Kaplan-Meier method and the statistical significance of different Materials and Methods groups was accessed by the log-rank test. The Cox’s regression pro- portional hazards model is applied to identify the independent prog- Patients and follow-ups nostic factors and the hazard ratio (HR) with 95% confidence interval This study was approved and supervised by the Ethics (95% CI) was used to evaluate the prognostic value. P-value < 0.05 Committee of the Affiliated Hospital of Shandong Medical College was considered statistically significant. and the Ethics Committee of Linyi People’s Hospital. There are total 353 patients who underwent surgery of EOC in the Affiliated Hospital of Shandong Medical College and the Linyi People’s Hospital from Results 2009 to 2013. Total of 89 patients were selected following the inclu- PGF expression is mainly in the cytoplasm of EOC tissues sion criteria: (1) the pathological type was serous ovarian cancer and The expression of PGF in serous ovarian cancer was the first operation was in advanced-stage (stage IIIB to IV), (2) there detected with IHC. In EOC tissues, PGF was mainly were available follow-ups and tissues for IHC, (3) there were no che- observed in the intracellular cytoplasm. According to the motherapies or radiotherapies before surgery, (4) patients got sys- cut-off, the cohort of our study was divided into the groups temic platinum-based chemotherapy after the operation, and (5) there of low PGF expression and high PGF expression (Fig. 1A, were no severe complications and other tumors occurred post-opera- B), which accounted for 61.8% (55/89) and 38.2% (34/89), tionally. The age of patients ranged from 41 to 73 years old, with the respectively. average age as 54.3 years old. The written informed consents were obtained from patients or their families. Optimal Debulking surgery was defined as a residual tumor no more than 1 cm in diameter PGF is associated with lymph node metastasis, TNM stage according to previous study (Zaid et al. 2013). The TNM stage of and response to chemotherapy patients was according to the AJCC/UICC staging system. The correlations between PGF expression and the clinicopathological factors were analyzed by the Fisher test Immunohistochemistry (Table 1). In our study, high expression of PGF was signifi- The paraffin-embedded specimens of serous EOC were sliced cantly associated with positive lymph node metastasis (P = into 4-μm sections for IHC according to previous studies (Xu et al. 0.004), indicating that high PGF may promote the progress 2014; Yang et al. 2014; Liu et al. 2017b). Briefly, 5% bovine serum of lymphatic invasion. Interestingly, patients with high albumin was used to incubate the specimen for 1 hour to block unspe- expression of PGF usually had poorer response to chemo- cific antigen binding after the endogenous peroxidase inactivation therapy (P = 0.001), suggesting that high PGF played a key antigen retrieval. The primary rabbit polyclonal PGF antibody (Cat role in the resistance of chemotherapy, which could help No. ab9542, Abcam, Cambridge, UK) at dilution 1:200 was applied stratify patients more preciously and explore the underlying overnight at 4℃, and the corresponding secondary antibody (Sangon mechanisms of drug resistance of EOC. Biotech, Shanghai, China) was used at room temperature for 2 hours after rinsing with phosphate buffered saline. The final signal visual- ization was achieved by application of 3,3′-diaminobenzidine. The PGF expression is predictive of unfavorable prognosis of slides were counterstained with hematoxylin finally. patients with EOC The relationship between the expression of PGF and Evaluation of IHC results the overall survival rate was evaluated by univariate analy- The results of IHC were evaluated and scored by two senior sis with Kaplan-Meier method and the difference between PGF is a Prognostic Biomarker in Advanced-Stage Serous Ovarian Cancer 293

Fig. 1. Representative immunohistochemical images of high- and low-expression of PGF in EOC. A. Image of low expression of PGF. The staining intensity was weak and the positive cells percentage was < 25%. The total IHC score was 1, and this case was defined as low expression of PGF. B. Image of high expression of PGF. The staining intensity was strong and the positive cells percentage was 50%-75%. The total IHC score was 9, and this case was defined as high expression of PGF.

Table 1. Correlation between PGF and the clinicopatholog- Independent prognostic risks of patients with EOC ical features. The prognostic factors were screened out by the uni- aate P variate analysis, and the independent prognostic risks were e g confirmed by multivariate analysis with the Cox-regression ge model (Table 2). The lymphatic status, TNM stage, . response to chemotherapy and expression of PGF were ≥ enrolled into the Cox-regression model. In our study, the ne etata positive lymph node metastasis and advanced TNM stage . were identified as the independent prognostic risks of EOC. e The poorer response to chemotherapy seemed to be the tlgal gae independent prognostic factors, with a statistical insignifi- . cant tendency (P = 0.053). However, the high-expression of PGF was not confirmed as an independent prognostic risk in our study (P = 0.518). This may be resulted from M tage that PGF had significant interaction with lymphatic metas- . tasis and response to chemotherapy, which were also enrolled into the multivariate analysis. PGF probably ene t etea induced unfavorable prognosis via promoting lymphatic . invasion or poor response to chemotherapy. e tetn tat Discussion tal . Combination of cytoreductive surgery and platinum- tal based therapy is the standard first-line therapy in the treat- *Fisher test. ment of EOC now (du Bois et al. 2003). However, ovarian PGF, placenta growth factor. cancer is a kind of malignancy with high heterogeneity, and epithelial serous ovarian cancer is the most common patho- the subgroups was analyzed with log-rank test (Table 2). logical type. Identification of prognostic biomarkers and High expression of PGF was demonstrated to be signifi- their mechanisms to affect prognosis at the molecular level cantly associated with lower overall survival rate of EOC (P could attribute to the drug target and targeted treatments. = 0.026) (Fig. 2A). The 5-year overall survival rates of But the exploration of biomarkers should be based on the patients with low and high expression of PGF were 37.9% separate pathological type, which is more important in and 10.2% respectively. In addition, the positive lymph ovarian cancer. PGF is a member of VEGF family and node metastasis (P = 0.005), advanced TNM stage (P = there are three VEGF receptors, which are VEGFR1 0.031), and poorer response to chemotherapy (P = 0.027) (FLT1), VEGFR2 (KDR, FLK1) and VEGFR3 (FLT4). were all remarkably correlated to the poorer prognosis (Fig. Among these receptors, PGF could only interact with 2B-D). VEGFR1, which is also a candidate biomarker in EOC (Kendall and Thomas 1993). We observed that high expres- 294 Q. Meng et al.

Table 2. Prognostic value of PGF and clincopathological features. Univariate analysis Multivariate analysis Characters 5-year survival rate% P* HR 95%CI P& Age < 50 40.5 ≥ 50 26.6 0.762 Lymph node metastasis No 48.6 1 Yes 6.2 0.005 2.59 1.29-5.18 0.007 Histological grade I 33.8 II 35.8 0.327 III 22.1 TNM stage III 36.0 1 IV 21.9 0.031 2.07 1.08-3.97 0.028 Response to chemotherapy Yes 37.8 1 No 30.6 0.027 2.32 0.99-5.46 0.053 Cytoreduction status optimal 37.0 suboptimal 9.1 0.947 PGF Low 37.9 1 High 10.2 0.026 1.27 0.62-2.60 0.518 *Calculated by Log-Rank test. &Calculated by Cox-regression model. PGF, placenta growth factor; HR, hazard ratio; CI, confidence interval.

Fig. 2. The correlation between PGF expression, lymphatic metastasis, TNM stage, response to chemotherapy and the survival rates. The overall survival curves were stratified by PGF expression (A), lymphatic metastasis (B), TNM stage (C) and response to chemotherapy (D). The statistical differences were calculated with the log-rank test. PGF is a Prognostic Biomarker in Advanced-Stage Serous Ovarian Cancer 295 sion of PGF was associated with the survival rate of cells through activation of mitogen-activated protein kinase patients with EOC in our study, without involving the (MAPK) signaling pathway. J. Biol. Chem., 289, 1415-1428. Burger, R.A. (2011) Overview of anti-angiogenic agents in devel- underlying molecular mechanisms, which has been opment for ovarian cancer. Gynecol. Oncol., 121, 230-238. explored previously (Song et al. 2015, 2016). In previous Cho, K.R. & Shih Ie, M. (2009) Ovarian cancer. Annu. Rev. study, VEGF-c was proved to promote ovarian cancer pro- Pathol., 4, 287-313. gression through paracrine and autocrine mechanisms Colombo, P.E., Fabbro, M., Theillet, C., Bibeau, F., Rouanet, P. & Ray-Coquard, I. (2014) Sensitivity and resistance to treatment (Decio et al. 2014). As a secretory growth factor, we also in the primary management of epithelial ovarian cancer. Crit. hypothesized that PGF may lead to the poorer prognosis in Rev. Oncol. Hematol., 89, 207-216. a paracrine or autocrine way synergetic with VEGFR1. But De Falco, S. (2012) The discovery of placenta growth factor and the hypothesis needs further validation with experiments its biological activity. Exp. Mol. Med., 44, 1-9. Decio, A., Taraboletti, G., Patton, V., Alzani, R., Perego, P., because PGF could also bind to neuropilin-1 and neuropi- Fruscio, R., Jurgensmeier, J.M., Giavazzi, R. & Belotti, D. lin-2 in a heparin-dependent manner. (2014) Vascular endothelial growth factor c promotes ovarian In our study, we demonstrated that high expression of carcinoma progression through paracrine and autocrine mech- PGF was significantly associated with the unfavorable anisms. Am. J. Pathol., 184, 1050-1061. prognosis in patients with advanced stage epithelial serous Dewerchin, M. & Carmeliet, P. (2014) Placental growth factor in cancer. Expert Opin. Ther. Targets, 18, 1339-1354. ovarian cancer, indicating that the anti-PGF therapy may be du Bois, A., Luck, H.J., Meier, W., Adams, H.P., Mobus, V., Costa, a potential treatment therapy to patients with EOC. S., Bauknecht, T., Richter, B., Warm, M., Schroder, W., However, it is interesting that multivariate analysis did not Olbricht, S., Nitz, U., Jackisch, C., Emons, G., Wagner, U., et define PGF expression as an independent prognostic factor al. (2003) A randomized clinical trial of cisplatin/paclitaxel versus carboplatin/paclitaxel as first-line treatment of ovarian in EOC, suggesting that high PGF expression may lead to cancer. J. Natl. Cancer Inst., 95, 1320-1329. poorer prognosis via influencing other factors, such as Fong, P.C., Yap, T.A., Boss, D.S., Carden, C.P., Mergui-Roelvink, lymph node metastasis and chemotherapy response. M., Gourley, C., De Greve, J., Lubinski, J., Shanley, S., Recently, many approaches have achieved significant pro- Messiou, C., A’Hern, R., Tutt, A., Ashworth, A., Stone, J., Carmichael, J., et al. (2010) Poly (ADP)-ribose polymerase gresses for the treatment of EOC. These approaches inhibition: frequent durable responses in BRCA carrier ovarian include proteins trapping, siRNA encapsulated in nanopar- cancer correlating with platinum-free interval. J. Clin. Oncol., ticles and humanized antibodies or specific small-molecular 28, 2512-2519. inhibitors (Landen et al. 2005; Yap et al. 2009). The most Jeruss, J.S. & Woodruff, T.K. (2009) Preservation of fertility in patients with cancer. N. Engl. J. Med., 360, 902-911. promising drug of EOC now is EGF antibody bevacizumab, Kendall, R.L. & Thomas, K.A. (1993) Inhibition of vascular endo- which is in a phase 3 clinical trial (Burger 2011). thelial cell growth factor activity by an endogenously encoded Moreover, PARP inhibitors are also potential targeted drugs soluble receptor. Proc. Natl. Acad. Sci. USA, 90, 10705- and are in pre-phase 3 now clinical trial (Fong et al. 2010). 10709. Khalil, A., Muttukrishna, S., Harrington, K. & Jauniaux, E. (2008) As to PGF signaling, there is an anti-PGF monoclonal anti- Effect of antihypertensive therapy with alpha methyldopa on body named RO5323441, which has been shown preclinical levels of angiogenic factors in pregnancies with hypertensive activity in several cancer models (Wang et al. 2017). The disorders. PLoS One, 3, e2766. pharmaceutical value of RO5323441 in EOC treatment Landen, C.N. Jr., Chavez-Reyes, A., Bucana, C., Schmandt, R., Deavers, M.T., Lopez-Berestein, G. & Sood, A.K. (2005) should be further investigated, but we considered Therapeutic EphA2 gene targeting in vivo using neutral RO5323441 as a potential targeted drug in EOC treatment liposomal small interfering RNA delivery. Cancer Res., 65, based on our clinical findings. 6910-6918. In summary, we detected the expression of PGF in 89 Liu, H., Xu, Y., Zhang, Q., Li, K., Wang, D., Li, S., Ning, S., Yang, cases of serous EOC and demonstrated that high expression H., Shi, W., Liu, Z. & Chen, Y. (2017a) Correlations between TBL1XR1 and recurrence of colorectal cancer. Sci. Rep., 7, of PGF is significantly associated with poorer response to 44275. chemotherapy and unfavorable prognosis of patients. Thus, Liu, H., Xu, Y., Zhang, Q., Yang, H., Shi, W., Liu, Z., Li, K., Gong, PGF is a prognostic biomarker, and detecting PGF could Z., Ning, S., Li, S. & Chen, Y. (2017b) Prognostic signifi- stratify the patients of high-risk and potential chemoresis- cance of TBL1XR1 in predicting liver metastasis for early stage colorectal cancer. Surg. Oncol., 26, 13-20. tance. The present findings also suggest the possibility that Rolny, C., Mazzone, M., Tugues, S., Laoui, D., Johansson, I., anti-PGF therapy is a promising therapy to treat EOC. Coulon, C., Squadrito, M.L., Segura, I., Li, X., Knevels, E., Costa, S., Vinckier, S., Dresselaer, T., Akerud, P., De Mol, M., et al. (2011) HRG inhibits tumor growth and metastasis by Conflict of Interest inducing macrophage polarization and vessel normalization The authors declare no conflict of interest. through downregulation of PlGF. Cancer Cell, 19, 31-44. Song, N., Liu, H., Ma, X. & Zhang, S. (2015) Placental growth factor promotes metastases of ovarian cancer through References MiR-543-regulated MMP7. Cell. Physiol. Biochem., 37, Barnes, M.N., Grizzle, W.E., Grubbs, C.J. & Partridge, E.E. (2002) 1104-1112. Paradigms for primary prevention of ovarian carcinoma. CA Song, N., Liu, H., Ma, X. & Zhang, S. (2016) Placental growth Cancer J. Clin., 52, 216-225. factor promotes ovarian cancer cell invasion via ZEB2. Cell. Basu, M., Mukhopadhyay, S., Chatterjee, U. & Roy, S.S. (2014) Physiol. Biochem., 38, 351-358. FGF16 promotes invasive behavior of SKOV-3 ovarian cancer Wang, K., Stark, F.S., Schlothauer, T., Lahr, A., Cosson, V., Zhi, J., 296 Q. Meng et al.

Habben, K., Tessier, J., Schick, E., Staack, R.F. & Krieter, O. Yang, X.Q., Xu, Y.F., Guo, S., Liu, Y., Ning, S.L., Lu, X.F., Yang, (2017) An apparent clinical pharmacokinetic drug-drug inter- H. & Chen, Y.X. (2014) Clinical significance of nerve growth action between bevacizumab and the anti-placental growth factor and tropomyosin-receptor-kinase signaling pathway in factor monoclonal antibody RO5323441 via a target-trapping intrahepatic cholangiocarcinoma. World J. Gastroenterol., 20, mechanism. Cancer Chemother. Pharmacol., 79, 661-671. 4076-4084. Xu, Y., Yang, X., Li, Z., Li, S., Guo, S., Ismail, S., Liu, H., Huang, Yap, T.A., Carden, C.P. & Kaye, S.B. (2009) Beyond chemo- Z., Zhang, Z., Chen, Y. & Sun, Q. (2017) Sprouty2 correlates therapy: targeted therapies in ovarian cancer. Nat. Rev. with favorable prognosis of gastric adenocarcinoma via Cancer, 9, 167-181. suppressing FGFR2-induced ERK phosphorylation and cancer Zaid, T.M., Yeung, T.L., Thompson, M.S., Leung, C.S., Harding, progression. Oncotarget, 8, 4888-4900. T., Co, N.N., Schmandt, R.S., Kwan, S.Y., Rodriguez-Aguay, Xu, Y.F., Yang, X.Q., Lu, X.F., Guo, S., Liu, Y., Iqbal, M., Ning, C., Lopez-Berestein, G., Sood, A.K., Wong, K.K., Birrer, M.J. S.L., Yang, H., Suo, N. & Chen, Y.X. (2014) Fibroblast & Mok, S.C. (2013) Identification of FGFR4 as a potential growth factor receptor 4 promotes progression and correlates therapeutic target for advanced-stage, high-grade serous to poor prognosis in cholangiocarcinoma. Biochem. Biophys. ovarian cancer. Clin. Cancer Res., 19, 809-820. Res. Commun., 446, 54-60.